Add Go API for TTS (#377)

.github/workflows/test-go-package.yaml

@@ -39,7 +39,7 @@ jobs:
           fetch-depth: 0
       - uses: actions/setup-go@v4
         with:
-          go-version: '>=1.20'
+          go-version: '>=1.12'
 
       - name: Display go version
         shell: bash
@@ -66,6 +66,121 @@ jobs:
         run: |
           gcc --version
 
+      - name: Test non-streaming TTS (Linux/macOS)
+        if: matrix.os != 'windows-latest'
+        shell: bash
+        run: |
+          mkdir tts-waves
+          cd go-api-examples/non-streaming-tts
+          ls -lh
+          go mod tidy
+          cat go.mod
+          go build
+          ls -lh
+
+          git lfs install
+
+          echo "Test vits-ljs"
+          git clone https://huggingface.co/csukuangfj/vits-ljs
+          ./run-vits-ljs.sh
+          rm -rf vits-ljs
+
+          echo "Test vits-vctk"
+          git clone https://huggingface.co/csukuangfj/vits-vctk
+          ./run-vits-vctk.sh
+          rm -rf vits-vctk
+
+          echo "Test vits-zh-aishell3"
+          git clone https://huggingface.co/csukuangfj/vits-zh-aishell3
+          ./run-vits-zh-aishell3.sh
+          rm -rf vits-zh-aishell3
+
+          ls -lh *.wav
+          cp *.wav ../../tts-waves/
+
+      - name: Test non-streaming TTS (Win64)
+        if: matrix.os == 'windows-latest' && matrix.arch == 'x64'
+        shell: bash
+        run: |
+          mkdir tts-waves
+          cd go-api-examples/non-streaming-tts
+          ls -lh
+          go mod tidy
+          cat go.mod
+          go build
+          ls -lh
+
+          echo $PWD
+          ls -lh /C/Users/runneradmin/go/pkg/mod/github.com/k2-fsa/
+          ls -lh /C/Users/runneradmin/go/pkg/mod/github.com/k2-fsa/*
+          cp -v /C/Users/runneradmin/go/pkg/mod/github.com/k2-fsa/sherpa-onnx-go-windows*/lib/x86_64-pc-windows-gnu/*.dll .
+          ls -lh
+
+          git lfs install
+
+          echo "Test vits-ljs"
+          git clone https://huggingface.co/csukuangfj/vits-ljs
+          ./run-vits-ljs.sh
+          rm -rf vits-ljs
+
+          echo "Test vits-vctk"
+          git clone https://huggingface.co/csukuangfj/vits-vctk
+          ./run-vits-vctk.sh
+          rm -rf vits-vctk
+
+          echo "Test vits-zh-aishell3"
+          git clone https://huggingface.co/csukuangfj/vits-zh-aishell3
+          ./run-vits-zh-aishell3.sh
+          rm -rf vits-zh-aishell3
+
+          ls -lh *.wav
+          cp *.wav ../../tts-waves/
+
+      - name: Test non-streaming TTS (Win32)
+        if: matrix.os == 'windows-latest' && matrix.arch == 'x86'
+        shell: bash
+        run: |
+          cd go-api-examples/non-streaming-tts
+          ls -lh
+          go mod tidy
+          cat go.mod
+          ls -lh
+
+          go env GOARCH
+          go env
+          echo "------------------------------"
+          go env -w GOARCH=386
+          go env -w CGO_ENABLED=1
+          go env
+
+          go clean
+          go build
+
+          echo $PWD
+          ls -lh /C/Users/runneradmin/go/pkg/mod/github.com/k2-fsa/
+          cp -v /C/Users/runneradmin/go/pkg/mod/github.com/k2-fsa/sherpa-onnx-go-windows*/lib/i686-pc-windows-gnu/*.dll .
+          ls -lh
+
+          git lfs install
+
+          echo "Test vits-ljs"
+          git clone https://huggingface.co/csukuangfj/vits-ljs
+          ./run-vits-ljs.sh
+          rm -rf vits-ljs
+
+          echo "Test vits-vctk"
+          git clone https://huggingface.co/csukuangfj/vits-vctk
+          ./run-vits-vctk.sh
+          rm -rf vits-vctk
+
+          echo "Test vits-zh-aishell3"
+          git clone https://huggingface.co/csukuangfj/vits-zh-aishell3
+          ./run-vits-zh-aishell3.sh
+          rm -rf vits-zh-aishell3
+
+          ls -lh *.wav
+          cp *.wav ../../tts-waves/
+
       - name: Test non-streaming decoding files (Linux/macOS)
         if: matrix.os != 'windows-latest'
         shell: bash
@@ -298,3 +413,8 @@ jobs:
           git clone https://huggingface.co/csukuangfj/sherpa-onnx-streaming-paraformer-bilingual-zh-en
           ./run-paraformer.sh
           rm -rf sherpa-onnx-streaming-paraformer-bilingual-zh-en
+
+      - uses: actions/upload-artifact@v3
+        with:
+          name: tts-waves
+          path: tts-waves

.github/workflows/test-go.yaml

@@ -60,6 +60,42 @@ jobs:
           go mod tidy
           go build
 
+      - name: Test non-streaming TTS (macOS)
+        shell: bash
+        run: |
+          mkdir tts-waves
+
+          cd scripts/go/_internal/non-streaming-tts/
+          ls -lh
+          go mod tidy
+          cat go.mod
+          go build
+          ls -lh
+
+          git lfs install
+
+          echo "Test vits-ljs"
+          git clone https://huggingface.co/csukuangfj/vits-ljs
+          ./run-vits-ljs.sh
+          rm -rf vits-ljs
+
+          echo "Test vits-vctk"
+          git clone https://huggingface.co/csukuangfj/vits-vctk
+          ./run-vits-vctk.sh
+          rm -rf vits-vctk
+
+          echo "Test vits-zh-aishell3"
+          git clone https://huggingface.co/csukuangfj/vits-zh-aishell3
+          ./run-vits-zh-aishell3.sh
+          rm -rf vits-zh-aishell3
+
+          cp *.wav ../../../../tts-waves/
+
+      - uses: actions/upload-artifact@v3
+        with:
+          name: tts-waves
+          path: tts-waves
+
       - name: Test non-streaming decoding files (macOS)
         shell: bash
         run: |

CMakeLists.txt

@@ -1,7 +1,7 @@
 cmake_minimum_required(VERSION 3.13 FATAL_ERROR)
 project(sherpa-onnx)
 
-set(SHERPA_ONNX_VERSION "1.8.3")
+set(SHERPA_ONNX_VERSION "1.8.4")
 
 # Disable warning about
 #

c-api-examples/offline-tts-c-api.c

@@ -188,7 +188,7 @@ int32_t main(int32_t argc, char *argv[]) {
   const SherpaOnnxGeneratedAudio *audio =
       SherpaOnnxOfflineTtsGenerate(tts, text, sid);
 
-  SherpaOnnxDestroyOfflineWriteWave(audio, filename);
+  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate, filename);
 
   SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
   SherpaOnnxDestroyOfflineTts(tts);

go-api-examples/non-streaming-decode-files/go.mod

@@ -1,9 +1,3 @@
 module non-streaming-decode-files
 
 go 1.12
-
-require (
-	github.com/k2-fsa/sherpa-onnx-go v1.7.12-alpha
-	github.com/spf13/pflag v1.0.5
-	github.com/youpy/go-wav v0.3.2
-)

go-api-examples/non-streaming-tts/go.mod

@@ -0,0 +1,3 @@
+module non-streaming-tts
+
+go 1.12

go-api-examples/non-streaming-tts/main.go

@@ -0,0 +1,61 @@
+package main
+
+import (
+	sherpa "github.com/k2-fsa/sherpa-onnx-go/sherpa_onnx"
+	flag "github.com/spf13/pflag"
+	"log"
+)
+
+func main() {
+	log.SetFlags(log.LstdFlags | log.Lmicroseconds)
+
+	config := sherpa.OfflineTtsConfig{}
+	sid := 0
+	filename := "./generated.wav"
+
+	flag.StringVar(&config.Model.Vits.Model, "vits-model", "", "Path to the vits ONNX model")
+	flag.StringVar(&config.Model.Vits.Lexicon, "vits-lexicon", "", "Path to lexicon.txt")
+	flag.StringVar(&config.Model.Vits.Tokens, "vits-tokens", "", "Path to tokens.txt")
+
+	flag.Float32Var(&config.Model.Vits.NoiseScale, "vits-noise-scale", 0.667, "noise_scale for VITS")
+	flag.Float32Var(&config.Model.Vits.NoiseScaleW, "vits-noise-scale-w", 0.8, "noise_scale_w for VITS")
+	flag.Float32Var(&config.Model.Vits.LengthScale, "vits-length-scale", 1.0, "length_scale for VITS. small -> faster in speech speed; large -> slower")
+
+	flag.IntVar(&config.Model.NumThreads, "num-threads", 1, "Number of threads for computing")
+	flag.IntVar(&config.Model.Debug, "debug", 0, "Whether to show debug message")
+	flag.StringVar(&config.Model.Provider, "provider", "cpu", "Provider to use")
+
+	flag.IntVar(&sid, "sid", 0, "Speaker ID. Used only for multi-speaker models")
+	flag.StringVar(&filename, "output-filename", "./generated.wav", "Filename to save the generated audio")
+
+	flag.Parse()
+
+	if len(flag.Args()) != 1 {
+		log.Fatalf("Please provide the text to generate audios")
+	}
+
+	text := flag.Arg(0)
+
+	log.Println("Input text:", text)
+	log.Println("Speaker ID:", sid)
+	log.Println("Output filename:", filename)
+
+	log.Println("Initializing model (may take several seconds)")
+
+	tts := sherpa.NewOfflineTts(&config)
+	defer sherpa.DeleteOfflineTts(tts)
+
+	log.Println("Model created!")
+
+	log.Println("Start generating!")
+
+	audio := tts.Generate(text, sid)
+
+	log.Println("Done!")
+
+	ok := audio.Save(filename)
+	if ok != 1 {
+		log.Fatalf("Failed to write", filename)
+	}
+
+}

go-api-examples/non-streaming-tts/run-vits-ljs.sh

@@ -0,0 +1,14 @@
+#!/usr/bin/env bash
+
+# please refer to
+# https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/vits.html#ljspeech-english-single-speaker
+# to download the model before you run this script
+
+./non-streaming-tts \
+  --vits-model=./vits-ljs/vits-ljs.onnx \
+  --vits-lexicon=./vits-ljs/lexicon.txt \
+  --vits-tokens=./vits-ljs/tokens.txt \
+  --sid=0 \
+  --debug=1 \
+  --output-filename=./vits-ljs.wav \
+  "Liliana, the most beautiful and lovely assistant of our team!"

go-api-examples/non-streaming-tts/run-vits-vctk.sh

@@ -0,0 +1,16 @@
+#!/usr/bin/env bash
+
+# please refer to
+# https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/vits.html#vctk-english-multi-speaker-109-speakers
+# to download the model before you run this script
+
+for sid in 0 10 108; do
+./non-streaming-tts \
+  --vits-model=./vits-vctk/vits-vctk.onnx \
+  --vits-lexicon=./vits-vctk/lexicon.txt \
+  --vits-tokens=./vits-vctk/tokens.txt \
+  --sid=0 \
+  --debug=1 \
+  --output-filename=./kennedy-$sid.wav \
+  'Ask not what your country can do for you; ask what you can do for your country.'
+done

go-api-examples/non-streaming-tts/run-vits-zh-aishell3.sh

@@ -0,0 +1,16 @@
+#!/usr/bin/env bash
+
+# please refer to
+# https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/vits.html#aishell3-chinese-multi-speaker-174-speakers
+# to download the model before you run this script
+
+for sid in 10 33 99; do
+./non-streaming-tts \
+  --vits-model=./vits-zh-aishell3/vits-aishell3.onnx \
+  --vits-lexicon=./vits-zh-aishell3/lexicon.txt \
+  --vits-tokens=./vits-zh-aishell3/tokens.txt \
+  --sid=10 \
+  --debug=1 \
+  --output-filename=./liliana-$sid.wav \
+  "林美丽最美丽、最漂亮、最可爱！"
+done

go-api-examples/real-time-speech-recognition-from-microphone/go.mod

@@ -1,9 +1,3 @@
 module real-time-speech-recognition-from-microphone
 
 go 1.12
-
-require (
-	github.com/gordonklaus/portaudio v0.0.0-20230709114228-aafa478834f5
-	github.com/k2-fsa/sherpa-onnx-go v1.7.12-alpha
-	github.com/spf13/pflag v1.0.5
-)

go-api-examples/streaming-decode-files/go.mod

@@ -1,9 +1,3 @@
 module streaming-decode-files
 
 go 1.12
-
-require (
-	github.com/k2-fsa/sherpa-onnx-go v1.7.12-alpha
-	github.com/spf13/pflag v1.0.5
-	github.com/youpy/go-wav v0.3.2
-)